This study compared the disfluencies of Japanese-speaking 24 preschool children who stutter (CWS) with those produced by age- and sex-matched comparison 24 children who do not stutter (CWNS). Part-word repetitions, prolongations and blocks were significantly more frequent in CWS than in CWNS, whereas no significant group differences occurred with respect to single-syllable word repetitions. Although a cut-off of 3 stuttering-like disfluencies (SLD) per 100 morae could classify only 46% of the CWS, 1 SLD per 100 morae could classify 96% of them. Our data showed single-syllable word repetitions are not the SLD and 1 SLD per 100 morae is an appropriate cut-off point in Japanese.